Recommender Systems for Learning

(1)

Nikos Manouselis, Hendrik Drachsler, Katrien

Verbert and Erik Duval

Recommender Systems for

Learning

– An introduction –

(2)

(3)

Preface

Recommender systems are extremely popular as a research and application area, with various interesting application domains such as e-commerce, entertainment, and others. Nevertheless, it was only around early 2000 when the first notable ap-plications appeared in the domain of education, since relevant work was generally considered to be connected to the area of adaptive educational systems.

Today, research around recommender systems in an educational context has sig-nificantly increased. Responding to this growing interest, this book expands the relevant chapter on Recommender Systems in Technology Enhanced Learning (by Manouselis, Drachsler, Vuorikari, Hummel and Koper) that was published in the Springer Recommender Systems Handbook (2011) and provides an extensive and in depth analysis of the recommender systems currently found in relevant literature. This book aims to briefly introduce recommender systems and to discuss a wide and representative sample of issues that people working on recommender systems for learning should be expecting to face. It serves as an overview of work in this domain and therefore especially addresses people that start studying or researching relevant topics and want to position their work in the overall landscape.

All the bibliography covered by this book is available in an open group created at the Mendeley research platform1and will continue to be enriched with additional references. We would like to encourage the reader to sign up for this group and to connect to the community of people working on these topics, gaining access to the collected blibliography but also contributing pointers to new relevant publications within this very fast emerging domain.

We hope that you will enjoy reading this book as much as we enjoyed working on it.

Athens, Heerlen, & Leuven Nikos Manouselis March, 2012 Hendrik Drachsler Katrien Verbert Erik Duval

(4)

(5)

Acknowledgements

We would like to thank all the people that have inspired this work and contributed to the dialogue around these topics in various fora and opportunities. Particular ap-preciation is reserved for our co-authors of the chapter upon which this work has developed and expanded: Riina Vuorikari, Hans Hummel and Rob Koper. We also thank Laura Gavrilut for her support in proof-reading the final version of the book.

The work presented in this book has been carried out with European Com-mission funding support. More specifically, the work of Nikos Manouselis has been supported by the EU project VOA3R - 250525 of the CIP PSP Programme (http://voa3r.eu). The work of Hendrik Drachsler was funded by the NeLLL funding body in the context of the AlterEgo project. Katrien Verbert is a Postdoctoral Fellow of the Research Foundation - Flanders (FWO). Part of this work was also supported by the SIG dataTEL of the European Association of Technology Enhanced Learn-ing and the former dataTEL Theme Team of the STELLAR Network of Excellence (grant agreement no. 231913). This publication reflects the views only of the au-thors, and the Commission cannot be held responsible for any use that may be made of the information contained therein.

(6)

(7)

Acronyms

AEH Adaptive Educational Hypermedia CAM Contextualised Attention Metadata

CSCL Computer Supported Collaborative Learning EDM Educational Data Mining

IR Information Retrieval ITS Intelligent Tutoring System KSA Knowledge, Skills, Abilities LAK Learning and Knowledge Analytics LMS Learning Management System LOM Learning Object Metadata MCDM Multi-Criteria Decision Making

MUPPLE Mash-Up Personal Learning Environment OAI Open Archives Initiative

PSLC Pittsburgh Science of Learning Center TEL Technology Enhanced Learning VLE Virtual Learning Environments

(10)

(11)

Chapter 1 Introduction and Background

Abstract In this chapter, we start with a short introduction to the increase that has been witnessed in the past few years in applications of recommender systems at the TEL domain. Then we provide some background on the area of recommender sys-tems, by defining recommender systems and outlining their basic types. A compari-son with relevant work in TEL is tried, particularly focusing on adaptive educational hypermedia, learning networks, educational data mining, and learning analytics. A discussion on their similarities and differences is also made, so that relevant work can be better positioned in the TEL research landscape.

1.1 Introduction

Technology enhanced learning (TEL) aims to design, develop and test sociotechni-cal innovations that will support and enhance learning practices of both individuals and organisations. It is therefore an application domain that generally covers tech-nologies that support all forms of teaching and learning activities. Since information retrieval (in terms of searching for relevant learning resources to support teachers or learners) is a pivotal activity in TEL, the deployment of recommender systems has attracted increased interest.

This should be more or less expected since a traditional problem in TEL has been the better findability of (mainly) digital learning resources. For instance, digital learning content is being regularly produced, organised and published in different types of TEL environments such as (Ochoa 2011):

1. Learning Object Repositories like Learning Resource Exchange1, Connexions2 or Maricopa Ex-change3;

1_{http://lreforschools.eun.org} 2_{http://cnx.org}

(12)

2 1 Introduction and Background

2. Learning Object Referratories like MERLOT4, OER Commons5or GLOBE6; 3. Open Courseware sites like MIT OCW7or OpenLearn8;

4. Learning Management Systems and Course Management Systems like Black-board9_{and Moodle}10_.

Various opportunities emerge for users to be exposed to this plethora of digital learning resources, in closed communities or in public and in both formal and non-formal settings. Potentially all user groups of TEL systems would find attractive services that help them identify suitable learning resources from this overwhelm-ing variety of choices. As a consequence, the concept of recommender systems be-came extremely appealing for TEL research. This is also reflected in the increasing networking and publication activities of researchers working on such applications. Recent examples include the Workshop series of Social Information Retrieval for Technology Enhanced Learning (SIRTEL 2007-2009), the RecSysTEL Workshop on Recommender Systems for Technology Enhanced Learning (Manouselis et al. 2010), the 1st dataTEL workshop on data sets for Technology Enhanced Learning (Drachsler et al. 2010; 2011; to appear), and several relevant special volumes of journals and books (Vuorikari et al. 2009; Verbert et al. 2010; Santos and Boticario 2012; Santos and Boticario in press; Tang et al. to appear). These efforts resulted a number of interesting observations, the main ones being that:

a) There is a large number of recommender systems that have been deployed (or that are currently under deployment) in TEL settings;

b) The information retrieval goals that TEL recommenders try to achieve are often different to the ones identified in other systems (e.g. product recommenders); c) There is a need to identify the particularities of TEL recommender systems, in

order to elaborate on methods for their systematic design, development and eval-uation.

Attempting to explore such particularities of this application domain, our book extends the analysis of Manouselis et al. (2011) in order to make a somewhat com-prehensive introduction of how recommender systems are deployed in TEL settings. Its main contribution is that it discusses a wide and representative set of issues that people working on recommender systems for learning should be expecting to face. It does not serve as an exhaustive review and analysis of available approaches and systems, but gives a rather fair overview of work in this domain.

The remainder of this book is structured as it follows. This chapter introduces recommender systems and discusses their relevance to similar areas in TEL. Chap-ter 2 focuses more on describing TEL as a recommendation context, defining the

4_{http://www.merlot.org} 5_{http://www.oercommons.org} 6_{http://globe-info.org} 7_{http://ocw.mit.edu} 8_{http://openlearn.open.ac.uk/} 9_{http://www.blackboard.com} 10_{http://moodle.org}

(13)

1.2 Recommender Systems 3

TEL recommendation problem and identifying relevant goals, supported user tasks, and variables of the TEL context that can be considered when making a recommen-dation. It also reviews data sets that are currently available from TEL applications and discusses how they could be useful for research on TEL recommender systems. Chapter 3 offers a comprehensive analysis of 42 recommender systems that have been found in relevant literature. Chapter 4 describes current challenges in the field and gives an outlook of future research trends in TEL recommender systems.

1.2 Recommender Systems

1.2.1 Definitions

Malone et al. (1987) provided an overview of intelligent information sharing sys-tems, referring to a fundamental categorisation of systems that generally support access to highly dynamic information resources (Belkin and Croft 1992; Baudisch 2001; Hanani et al. 2001). More specifically, they distinguished cognitive filtering systems as the ones that characterise the contents of an information resource (shortly referred to as an item) and the information needs of potential item users, and then use these representations to intelligently match items to users; and sociological fil-tering systems as the ones that are working based on the personal and organisational interrelationships of individuals in a community. Early information sharing systems belonged to the first category and were based on text-based filtering, which works by selecting relevant items according to aset of textual keywords (Konstan 2004). Collaborative filtering systems were first introduced as representatives of the second category. They addressed two problems of text-based systems:

• The problem of overwhelming numbers of on-topic items (ones which would be all selected by a keyword filter), which has been addressed by the introduction of further evaluating the items based on human judgment about their quality. • The problem of filtering non-text items, which has been addressed by judging

them solely upon human taste.

Therefore, early recommender systems were based on the notion of collaborative filtering, and have been defined as systems that ”...help people make choices based on the opinions of other people.” (Goldberg et al. 1992). As years came by, the term ”recommender systems” has prevailed over the term ”collaborative filtering systems”. It first described systems in which ”...people provide recommendations as inputs, which the system then aggregates and directs to appropriate recipients.” (Resnick and Varian 1997). Finally, it evolved to a meaning that is more or less valid today, covering ”...any system that produces individualised recommendations as output or has the effect of guiding the user in a personalised way to interest-ing or useful objects in a large space of possible options.” (Burke 2002; Burke and

(14)

from text-based ones: the criterion of ’individualised’ and the criterion of ’inter-esting and useful’ content. Table 1.1 provides an overview of relevant definitions that we have identified in the literature, extending the initial collection reported in Manouselis and Costopoulou (2007).

1.2.2 Types

In the literature, recommender systems have been usually classified into two basic types, according to the way recommendations are made (Adomavicius and Tuzhilin 2005):

• Content-based recommendation, in which the user is recommended items similar to the ones he has preferred in the past. Content-based recommendation systems analyse a set of items and/or descriptions previously preferred by a user, and build a model or profile of user interests based on the features of these items (Lops et al. 2011; Pazzani and Bilsus 1997).

• Collaborative recommendation, in which the user is recommended items that people with similar tastes and preferences liked in the past. Collaborative rec-ommendation (or collaborative filtering) systems predict a user’s interest in new items based on the recommendations of other people with similar interests (Schafer et al. 2007; Ekstrand et al. 2010).

Moreover, other types of recommender systems have been also proposed in the liter-ature. For instance, Burke (2002; 2007) distinguishes the following ones (in addition to the two described above):

• Demographic recommendation, which classifies the users according to the at-tributes of their personal profile, and makes recommendations based on demo-graphic classes.

• Utility-based recommendation, which makes suggestions based on a computation of the utility of each item for a user, for whom a utility function has to be stored. • Knowledge-based recommendation, which suggests items based on logical infer-ences about user preferinfer-ences. A knowledge representation (e.g. rules) about how an item meets a particular user need is necessary.

Furthermore, Adomavicius and Tuzhilin (2005) also distinguish recommenders in those that aim to predict absolute values of ratings users would give to yet un-seen items, from preference-based filtering, i.e. predicting the relative preferences of users. Finally, hybrid recommendation has also been identified. Recommender systems of this type combine two or more of the aforementioned types in order to gain better performance and address the shortcomings of each type (Burke 2002; 2007).

(15)

1.2 Recommender Systems 5

Table 1.1: Overview of definitions related to recommender systems.

Goldberg et al. 1992

”Collaborative filtering simply means that people collaborate to help one another perform filtering by recording their reactions to documents they read.”

Resnick et al. 1994

”Collaborative filters help people make choices based on the opinions of other peo-ple.”

Shardanand and Maes 1995

”Social information filtering essentially automates the process of ’word-of-mouth’ recommendations: items are recommended to a user based upon values assigned by other people with similar taste.”

Resnick and

Varian 1997

”In a typical recommender system people provide recommendations as inputs, which the system then aggregates and directs to appropriate recipients.”

Pennock and

Horvitz 1999; Goldberg et al. 2001

”The term ’collaborative filtering’ describes techniques that use the known prefer-ences of a group of users to predict the unknown preferprefer-ences of a new user; recom-mendations for the new users are based on these predictions. Other terms that have been proposed are ’social information filtering’ and ’recommender system’.” Schafer et al.

2001

”Recommender systems use product knowledge -either hand-coded knowledge pro-vided by experts or ’mined’ knowledge learned from the behavior of consumers- to guide consumers through the often-overwhelming task of locating products they will like.”

Burke 2002;

Lops et al. 2011

”...any system that produces individualised recommendations as output or has the effect of guiding the user in a personalized way to interesting or useful objects in a large space of possible options.”

Konstan 2004 ”Recommender systems use the opinions of members of a community to help in-dividuals in that community identify the information or products most likely to be interesting to them or relevant to their needs.”

Herlocker et al. 2004

”Recommender systems use the opinions of a community of users to help individuals in that community more effectively identify content of interest from a potentially overwhelmingset of choices.”

Deshpande and Karypis 2004

”Recommender systems - a personalized information filtering technology used to either predict whether a particular user will like a particular item (prediction prob-lem) or to identify aset of N items that will be of interest to a certain user (top-N recommendation problem).”

Hung 2005 ”A personalized recommendation system can provide one-to-one service to cus-tomers based on cuscus-tomers’ past behavior and through inference from other users with similar preferences. The aim of personalization is to offer customers what they want without asking explicitly and to capture the social component of interpersonal interaction.”

Schein et al. 2005

”Recommender systems suggest items of interest to users based on their explicit and implicit preferences, the preferences of other users, and user and item attributes.” Smyth 2007 ”Recommender systems try to help users access complex information spaces.” Burke 2007 ”Recommender systems are personalized information agents that provide

recom-mendations: suggestions for items likely to be of use to a user... A recommender can be distinguished from an information retrieval system by the semantics of its user interaction.”

(16)

1.3 Relevant Systems in Educational Applications

1.3.1 Adaptive Educational Hypermedia

Web systems generally suffer from the inability to satisfy the heterogeneous needs of many users. To address this challenge, a particular strand of research that has been called adaptive web systems (or adaptive hypermedia) tried to overcome the short-comings of traditional ’one-size-fits-all’ approaches by exploring ways in which Web-based could adapt their behaviour to the goals, tasks, interests, and other char-acteristics of interested users (Brusilovsky and Nejdl 2004). A particular category of adaptive systems has been the one dealing with educational applications, called adaptive educational hypermedia (AEH) systems.

Adaptive web systems belong to the class of user-adaptive software systems (Schneider-Hufschmidt et al. 1993). According to Oppermann (1994) a system is called adaptive ”if it is able to change its own characteristics automatically accord-ing to the user’s needs”. Adaptive systems consider the way the user interacts with the system and modify the interface presentation or the system behaviour accord-ingly (Weibenzahl 2003). Jameson (2001) adds an important characteristic: a user-adaptive system is an interactive system which adapts its behaviour to each individ-ual user on the basis of nontrivial inferences from information about that user.

Adaptive systems help users find relevant items in a usually large information space, by essentially engaging three main adaptation technologies (Brusilovsky and Nejdl 2004): adaptive content selection, adaptive navigation support, and adaptive presentation. The first of these three technologies comes from the field of adaptive information retrieval (IR) (Baudisch 2001) and is associated with a search-based access to information. When the user searches for relevant information, the system can adaptively select and prioritise the most relevant items. The second technology was introduced by adaptive hypermedia systems (Brusilovsky 1996) and is associ-ated with a browsing-based access to information. When the user navigates from one item to another, the system can manipulate the links (e.g., hide, sort, annotate) to guide the user adaptively to most relevant information items. The third technol-ogy has its roots in the research on adaptive explanation and adaptive presentation in intelligent systems (Moore and Swartout 1990; Paris 1988). It deals with presen-tation, not access to information. When the user gets to a particular page, the system can present its content adaptively.

As Brusilovksy (2001) describes, educational hypermedia was one of the first application areas of adaptive systems. A simplified architecture of the layers within an educational AEH system that has been developed simplifying the elaborate one found in Karampiperis and Sampson (2005) is presented in Figure 1.1. This archi-tecture includes: a layer including the representation and organisation of knowledge about educational content (learning resources), the domain (domain ontology), and the user (user model); a layer that includes the adaptation mechanisms and rules; and a layer that provides the run-time adaptation results to the user. A number of pi-oneer adaptive educational hypermedia systems were developed between 1990 and

(17)

1.3 Relevant Systems in Educational Applications 7

Fig. 1.1: Generic layers within a simplified example architecture of an educational AEH (adapted from: Karampiperis and Sampson 2005; Manouselis et al. 2011)

1996, which Brusilovksy roughly divided into two research streams. The first stream includes systems created by researchers in the area of intelligent tutoring systems (ITS) who were trying to extend traditional student modelling and adaptation ap-proaches developed in this field to ITS with hypermedia components (Brusilovsky et al. 1993; Gonschorek and Herzog 1995; Prez et al. 1995). The systems of the second stream were developed by researchers working on educational hypermedia in an attempt to make their systems adapt to individual students (De Bra 1996; De La Passardiere and Dufresne 1992; Hohl et al. 1996; Kay and Kummerfeld 1994).

AEH research has often followed a top-down approach, greatly depending on expert knowledge and involvement in order to identify and model TEL context vari-ables. For example, Cristea (2005) describes a number of expertise-demanding tasks when AEH content is authored: initially creating the resources, labeling them, com-bining them into what is known as a domain model; then, constructing and main-taining the user model in a static or dynamic way, since it is crucial for achiev-ing successful adaptation in AEH. Generally speakachiev-ing, in AEH a large amount of user-related information (characterising needs and desires) has to be encoded in the content creation phase. This can take place in formal educational settings when the context variables are usually known, and there is significant AEH research (e.g.

(18)

1.3.2 Learning Networks

Another strand of work includes research where the context variables are extracted from the contributions of the users. A category of such systems includes learn-ing networks, which connect distributed learners and providers in certain domains (Koper and Tattersall 2004; Koper et al. 2005). The design and development of learning networks is highly flexible, learner-centric and evolving from the bottom upwards, going beyond formal course and programme-centric models that are im-posed from the top downwards. A learning network is populated with many learners and learning activities provided by different stakeholders. Each user is allowed to add, edit, delete or evaluate learning resources at any time.

The concept of learning networks (Koper et al. 2005) provides methods and tech-nical infrastructures for distributed lifelong learners to support their personal com-petence development. It takes advantages of the possibilities of the Web 2.0 devel-opments and describes the new dynamics of learning in the networked knowledge society. A learning network is learner-centered and its development emerges from the bottom-up through the participation of the learners. Emergence is the central idea of the learning network concept. Emergence appears when an interacting sys-tem of individual actors and resources self-organises to shape higher-level patterns of behaviour (Gordon 1999; Johnson 2001; Waldrop 1992).

We can imagine users (e.g. learners) interacting with learning activities in a learn-ing network while their progress is belearn-ing recorded. Indirect measures like time or learning outcomes, and direct measures like ratings and tags given by users allow to identify paths in a learning network which are faster to complete or more attractive than others (e.g. Drachsler et al. 2009a; Vuorikari and Koper 2009). This informa-tion can be fed back to other learners in the learning network, providing collective knowledge of the ’swarm of learners’ in the learning network. Most learning en-vironments are designed only top-down as often times their structure, learning ac-tivities and learning routes are predefined by an educational institution. Learning networks, on the other hand, take advantage of the user-generated content that is created, shared, rated and adjusted by using Web 2.0 technologies. In the field of TEL, several European projects address these bottom-up approaches of creating and sharing knowledge, such as the TENcompetence project (Wilson et al. 2008) or the LTfLL project (Drachsler et al. 2010).

Following a similar approach, in Research Networks researchers are intercon-nected over Web 2.0 tools and are informed about latest research activities. This combined information of a specific research community easily becomes overwhelm-ing, thus also researchers face an information overflow issue. Customised awareness support tools are needed to visualise and explore the collected data. But also recom-mender systems are becoming increasingly important to support researchers in the daily work process (Reinhardt et al. 2011a; Reinhardt et al. 2011b).

Another category of systems that formulate and define their context variables following a bottom-up approach, are Mash-Up Personal Learning Environments (MUPPLE) (Wild et al. 2008). First such approaches were created by (Liber 2000; Liber and Johnson 2008; Wild et al. 2008). The iCamp EU-initiative explicitly

(19)

ad-1.3 Relevant Systems in Educational Applications 9

Fig. 1.2: Evolution of a learning network from Drachsler (2009b) (left (A): starting phase with a first learner moving through possible learning activities; right (B): advanced phase showing emerging learning paths from the collective behavior of all learners)

dresses the integration of Web 2.0 sources into MUPPLE, by creating a flexible environment that allows learners to create their own environments for certain learn-ing activities. MUPPLEs are a kind of instance of the learnlearn-ing network concept and therefore share several characteristics with it. They also support informal learn-ing as they require no institutional background and focus on the learner instead of institutional needs like student management or assessments. The learners do not par-ticipate in formal courses and neither receive any certification for their competence development. A common problem for MUPPLEs is the amount of data that is gath-ered already in a short time frame and the unstructured way it is collected. This can make the process of user and domain modelling demanding and unstructured. On the other hand, this is often the case in recommender systems as well, when user and item interactions are explored, e.g. in order to identify user and item similarities.

1.3.3 Educational Data Mining and Learning Analytics

Educational Data Mining (EDM) is an emerging discipline that has attracted sig-nificant interest during the past years (Romero et al. 2008; Baker 2010). This in-terest leads into the creation of a relevant scientific society (International EDM so-ciety), a dedicated journal (Journal of EDM) and an annual conference that has already reached its fifth edition (http://www.educationaldatamining.org). By defini-tion (Baker 2010), EDM explores the applicadefini-tion of data mining methods in order

(20)

methodologies followed to analyse educational data, ways to integrate data min-ing with pedagogical theories, as well as applications that are used for improvmin-ing educational software or teacher support, for improving understanding of learners’ domain representations, and for improving assessment of learners’ engagement in the learning tasks.

Traditional data mining methods are used to support mining educational data sets, but trying to discover and take advantage of the unique features of educational data. Baker and Yacef (2010) classified the EDM areas into: prediction (e.g. classification, regression, density estimation); clustering; relationship mining (e.g. association rule mining, correlation mining, sequential pattern mining, causal data mining); distilla-tion of data for human judgment; and discovery with models. In EDM, whether educational data is taken from students’ use of interactive learning environments, computer-supported collaborative learning, or administrative data from schools and universities, it often has multiple levels of meaningful hierarchy, which often need to be determined by properties in the data itself, rather than in advance. Furthermore, issues of time, sequence, and context also play important roles in the study of educa-tional data. The work in this area can be considered to be relevant to the domain of recommender systems for educational applications, since many recommender sys-tems apply data mining techniques in order to cluster users, find correlations and improve their recommendations (Romero and Ventura 2010).

In addition, an emerging strand of research is around the topic of the so-called Learning and Knowledge Analytics (LAK), as reflected by a number of confer-ences and special issues in recent years (Siemens 2010; Siemens and Gasevic 2011). Among others, the analysis of learner data and identification of patterns within these data are researched to predict learning outcomes, to suggest relevant resources and to detect error patterns or affects of learners. The definition of Learning Analytics by Siemens (2010) describes them as the use of intelligent data, learner-produced data, and analysis models to discover information and social connections, and to predict and advise on learning. This definition reveals that Learning Analytics are very closely related to EDM, with a particular emphasis on knowledge representa-tion and reasoning (Romero and Ventura 2007). This is perfectly justified since, in an increasing number of scientific disciplines, large data collections are emerging as important community resources (Chervenak et al. 2000). These data sets are used as benchmarks to develop new algorithms and compare them to other algorithms in given settings. For instance, when data sets are intended to be used for recommenda-tions algorithms, various data types such as explicit (ratings) or implicit (downloads and tags) can serve as potential relevance indicators.

1.3.4 Similarities and differences

Many of the AEH systems address formal learning (e.g. Aroyo et al. 2003; De Bra et al. 2002; Kravcik et al. 2004), have equally fine granulated knowledge domains and can therefore offer personalised recommendations to the learners. They take

(21)

1.3 Relevant Systems in Educational Applications 11

advantage of technologies like metadata and ontologies to define the relationships, conditions, and dependencies of learning resources and learner models. These sys-tems are mainly used in ’closed-corpus’ applications (Brusilovsky and Henze 2007) where the learning resources can be described by an educational designer through semantic relationships and is therefore a formal learning offer. As mentioned be-fore, in formal educational settings (such as universities) there are usually well-structured formal relationships like predefined learning plans (curriculum) with lo-cations, student/teacher profiles, and accreditation procedures. All this metadata can be used to recommend courses or personalise learning through the adaptation of the learning resources or the learning environment to the students (Baldoni et al. 2007). One interesting direction in this research is the work on adaptive sequencing which takes into account individual characteristics and preferences for sequencing learn-ing resources (Karampiperis and Sampson 2005). In AEH there are many design activities needed before the runtime and also during the maintenance of the learn-ing environment. In addition, the knowledge domains in the learnlearn-ing environment need to be described in detail. These aspects make adaptive sequencing and other adaptive hypermedia techniques less applicable for TEL recommendation, where informal learning networks emerge without some highly structured domain model representation.

In informal learning networks, mining techniques need to be used in order to cre-ate some representation of the user or domain model. For instance, prior knowledge in informal learning is a rather diffuse parameter because it relies on information given by the learners without any standardisation. To handle the dynamic and dif-fuse characteristic of prior knowledge, and to bridge the absence of a knowledge domain model, probabilistic techniques like latent semantic analysis are promising (van Bruggen et al. 2004). The absence of maintenance and structure in informal learning is also called the ’open corpus problem’. The open corpus problem applies when an unlimited set of documents is given that cannot be manually structured and indexed with domain concepts and metadata from a community (Brusilovsky and Henze 2007). The open corpus problem also applies to informal learning net-works. Therefore, bottom-up recommendation techniques like collaborative filter-ing are more appropriate because they require nearly no maintenance and improve through the emergent behaviour of the community. Drachsler et al. (2008b) anal-ysed how various types of collaborative filtering techniques can be used to support learners in informal learning networks. Following their conclusions we have to con-sider the different environmental conditions of informal learning, such as the lack of maintenance and less formal structured learning objects, in order to provide an appropriated navigation support to recommender systems. Learning networks are mainly structured by tags and ratings given by their users, being therefore in con-trast with the institutionalised Learning Management Systems (LMSs) like Moo-dle (http://mooMoo-dle.org) or Blackboard (http://www.blackboard.com) that are used to better manage learning activities and distribute learning resources to learners.

(22)

learn-12 1 Introduction and Background

Table 1.2: A selected list of recommendation techniques used in TEL and their use-fulness for learning. Extend version based on initial table of Drachsler et al. (2008a).

usually very small compared to the number of ratings that have to be predicted. Effective prediction by ratings based on small amounts is very essential for rec-ommender systems and has an effect on the selection of a specific

(23)

recommenda-References 13

tion technique. Formal learning can rely on regular evaluations of experts or stu-dents upon multiple criteria (e.g., pedagogical quality, technical quality, ease of use) (Manouselis et al. 2007), but in informal learning environments such evalu-ation procedures are unstructured and few. Formal learning environments like uni-versities often have integrated evaluation procedures for a regular quality evaluation to report to their funding body. With these integrated evaluation procedures more dense data sets can be expected. As a conclusion, the data sets in informal learn-ing context are characterised by the ’Sparsity problem’ caused by sparse ratlearn-ings in the data set. Multi-criteria ratings could be beneficial for informal learning to over-come the ’Sparsity problem’ of the data sets. These multi-criteria ratings have to be reasonable for the community of lifelong learners. The community could rate learn-ing resources on various levels, such as required prior knowledge level (novice to expert), the presentation style of learning resources, and even the level of attractive-ness, because keeping students satisfied and motivated is a vital criteria in informal learning. These explicit rating procedures should be supported with several indirect measures like ’Amount of learners using the learning resource’ or ’Amount of ad-justments of a learning resources’, in order to measure how up-to-date the learning resource is.

Informal learning is therefore different from well-structured domains, like for-mal learning. Recommender systems for inforfor-mal learning have no official main-tenance by an institution, mostly rely on its community and most of the time do not contain well-defined metadata structures. Moreover, formal learning is charac-teristically top-down designed and the learning contents are a closed-corpus that can only be edited by domain experts; Informal learning contents emerge from the bottom-upwards through communities contributions (open-corpus) and every com-munity member can adjust and contribute information. Therefore, it will be difficult to transfer and apply recommender systems even from formal to non-formal settings (and vice-versa), since user tasks and recommendation goals are often substantially different.

A critical assessment of recommender techniques regarding their applicability and usefulness in TEL has taken place by Drachsler et al. (2008a). Table 1.2 pro-vides an initial overview of advantages and disadvantages of each technique, and reports the envisaged usefulness of each technique for TEL recommenders. Never-theless, it aims to serve as an initial basis for such a discussion, since a more detailed and elaborate survey of all existing recommendation methods and techniques can take place in the future.

References

(24)

Aroyo L, Mizoguchi R, Tzolov C (2003) OntoAIMS: Ontological Approach to Courseware Authoring. Paper presented at the conference International Confer-ence on Computers in Education (ICCE 2003), 2-5 December 2003, Hong Kong, China, p. 8.

Baker R, Yacef K (2010) The State of Educational Data Mining in 2009: A Review and Future Visions. Journal of Educational Data Mining, 1(1):3-17.

Baker R. (2010) Data Mining for Education. In: McGaw B, Peterson P, Baker E (eds) International Encyclopedia of Education (3rd edition), 7:112-118. Oxford, UK: Elsevier.

Baldoni M, Baroglio C, Brunkhorst I, Marengo E, Patti V (2007) Reasoning-Based Curriculum Sequencing and Validation: Integration in a Service-Oriented Archi-tecture. In: Duval E, Klamma R, Wolpers M (eds) Proceedings of the 2nd Euro-pean Conference on Technology Enhanced Learning, LNCS 4753, pp. 426-431. Berlin; Heidelberg; New York: Springer.

Baudisch P (2001) Dynamic Information Filtering. PhD Thesis, GMD Research Series, No 16, Darmstad Technical University, p. 188.

Belkin NJ, Croft WB (1992) Information Filtering and Information Retrieval: Two Sides of the Same Coin? Communications of the ACM, 35(12):29-38.

Brusilovsky P, Nejdl W (2004) Adaptive Hypermedia and Adaptive Web. Practical Handbook of Internet Computing, CRC Press LLC.

Brusilovsky P (1996) Methods and techniques of adaptive hypermedia. User Mod-eling and User-Adapted Interaction, 6(2-3):87-129.

Brusilovsky P (2001) Adaptive hypermedia. User Modeling and User-Adapted In-teraction, 11(1-2):87-110.

Brusilovsky P, Henze N (2007) Open Corpus Adaptive Educational Hypermedia. In Brusilovsky P, Kobsa A, Nejdl W (eds), The Adaptive Web: Methods and Strate-gies of Web Personalization, LNCS 4321, pp. 671-696. Berlin Heidelberg New York: Springer.

Brusilovsky P, Pesin L, Zyryanov M (1993) Towards an adaptive hypermedia com-ponent for an intelligent learning environment. In Bass LJ, Gornostaev J, Unger C (eds), Human-Computer Interaction, LNCS, Vol. 753, pp. 348-358. Berlin: Springer-Verlag.

Burke R (2002) Hybrid Recommender Systems: Survey and Experiments. User Modeling and User-Adapted Interaction, 12:331-370.

Burke R (2007) Hybrid Web Recommender Systems. In Brusilovsky P, Kobsa A, Neidl W (eds) The Adaptive Web: Methods and Strategies of Web Personaliza-tion, LNCS, Vol. 4321, Springer-Verlag, Berlin Heidelberg New York, 377-408. Burke R, Ramezani M (2011) Matching Recommendation Technologies and

Do-mains. In Ricci F, Rokach L, Shapira, B. (eds) Recommender Systems Handbook, pp. 367-386. Springer US.

Chervenak A, Foster I, Kesselman C, Salisbury C, Tuecke S (2000) The data grid: Towards an architecture for the distributed management and analysis of large sci-entific data sets. Journal of Network and Computer Applications, 23(3):187-200. Cristea A (2005) Authoring of Adaptive Hypermedia. Educational Technology and

(25)

References 15

De Bra P (1996) Teaching Hypertext and Hypermedia through the Web. Journal of Universal Computer Science, 2(12):797-804.

De Bra P, Aerts A, Smits D, Stash N (2002) AHA! Version 2.0, More Adaptation Flexibility for Authors. Paper presented at the conference World Conference on e-Learning in Corporate, Goverment, Healthcare & Higher Education. 15-19 Oc-tober 2002, Montreal, Canada.

De La Passardiere B, Dufresne A (1992) Adaptive navigational tools for educational hypermedia. In: Tomek I (ed), ICCAL’92, 4th International Conference on Com-puters and Learning, pp. 555-567. Berlin: Springer-Verlag.

Deshpande M, Karypis G (2004) Item-based Top-N Recommendation Algorithms. ACM Transactions on Information Systems, 22(1):143-177.

Drachsler H, Hoisl B, Wagenlecher C (2010). User-tailored Inter-Widget Commu-nication. Extending the Shared Data Interface for the Apache Wookie Engine, International Conference on Interactive Computer Aided Learning 2010, Hasselt, Belgium.

Drachsler H, Hummel H, Koper R (2008a) Personal recommender systems for learn-ers in lifelong learning: requirements, techniques and model. International Jour-nal of Learning Technology, 3(4):404-423.

Drachsler H, Bogers T, Vuorikari R, Verbert K, Duval E, Manouselis N, Beham G, Lindstaedt S, Stern H, Friedrich M (2010) Issues and considerations regarding sharable data sets for recommender systems in technology enhanced learning. Procedia Computer Science, 1(2):2849-2858. doi:10.1016/j.procs.2010.08.010 Drachsler H, Verbert K, Sicilia M-A, Wolpers M, Manouselis N, Vuorikari R,

Lind-staedt S, Fischer F (2011) dataTEL - Datasets for Technology Enhanced Learn-ing. STELLAR ARV White Paper. Available at: http://oa.stellarnet.eu/open-archive/browse?resource=6756 v1 [Accessed March 12, 2012].

Drachsler H, Hummel H, Koper R (2008b) Using Simulations to Evaluate the Ef-fects of Recommender Systems for Learners in Informal Learning Networks. In Vuorikari R, Kieslinger B, Klamma R, Duval E (eds) SIRTEL workshop at the 3rd EC-TEL conference. CEUR Workshop Proceedings VOL-382, Maastricht, The Netherlands.

Drachsler H, Hummel H, Van den Berg B, Eshuis J, Berlanga A, Nadolski R, Wa-terink W, Boers N, Koper R (2009a) Effects of the ISIS Recommender System for navigation support in self-organised learning networks. Journal of Educational Technology and Society, 12:122-135.

Drachsler H (2009b) Navigation Support for Learners in Informal Learn-ing Networks. Open University of the Netherlands. Available at: http://dspace.ou.nl/handle/1820/2084 [Accessed March 12, 2012].

Drachsler H, Verbert K, Manouselis N, Lindstaedt S, Vuorikari R, Wolpers, M (to appear). Preface for dataTEL Special Issue on Datasets and Data Supported Learning in Technology-Enhanced Learning. Drachsler H, Verbert K, Manouselis N, Lindstaedt S, Vuorikari R, Wolpers M (eds) International Journal of Technol-ogy Enhanced Learning (IJTEL).

(26)

Goldberg D, Nichols D, Oki BM, Terry D (1992) Using Collaborative Filtering to Weave an Information Tapestry. Communications of the ACM, 35(12):61-70. Goldberg K, Roeder T, Gupta D, Perkins C (2001) Eigentaste: A Constant Time

Collaborative Filtering Algorithm. Information Retrieval, 4(2):133-151.

Gonschorek M, Herzog C (1995) Using hypertext for an adaptive helpsystem in an intelligent tutoring system. In: Greer J (ed) Proceedings of the 7th World Con-ference on Artificial Intelligence in Education, pp. 274-282. Washington, DC, AACE.

Gordon D (1999) Ants at Work: How an Insect Society is Organized. Free Press, New York.

Hanani U, Shapira B, Shoval P (2001) Information Filtering: Overview of Issues, Research and Systems. User Modeling and User-Adapted Interaction, 11:203-259.

Herlocker JL, Konstan JA, Terveen LG, Riedl JT (2004) Evaluating Collaborative Filtering Recommender Systems. ACM Transactions on Information Systems, 22(1):5-53.

Hohl H, B¨ocker H-D, Gunzenh¨auser R (1996) Hypadapter: An adaptive hypertext system for exploratory learning and programming. User Modeling and User-Adapted Interaction, 6(2-3):131-156.

Hung L-P (2005) A personalized recommendation system based on product taxon-omy for one-to-one marketing online. Expert Systems with Applications, 29:383-392.

Jameson A (2001) Systems That Adapt to Their Users: An Integrative Perspective. Saarbr¨ucken: Saarland University.

Johnson S (2001) Emergence. Scribner, New York.

Karampiperis P, Sampson D (2005) Adaptive Learning Resources Sequencing in Educational Hypermedia Systems. Educational Technology and Society, 8(4):128-147.

Kay J, Kummerfeld RJ (1994) An individualised course for the C programming language. In: Proceedings of Second International WWW Conference, 17-20 Oc-tober, 1994. Chicago, USA.

Konstan JA (2004) Introduction To Recommender Systems: Algorithms and Evalu-ation. ACM Transactions on Information Systems, 22(1):1-4.

Koper R, Tattersall C (2004) New directions for lifelong learning using network technologies. British Journal of Educational Technology, 35(6):689-700. Koper R, Rusman E, Sloep P (2005) Effective Learning Networks. Lifelong

Learn-ing in Europe, 1:18-27.

Kravcik M, Specht M, Oppermann R (2004) Evaluation of WINDS Authoring En-vironment. In: De Bra P, Nejdl W (eds) Adaptive Hypermedia and Adaptive Web-Based Systems, LNCS, Vol. 3137, pp. 166-175. Berlin: Springer.

Liber O (2000) Colloquia - a conversation manager. Campus-Wide Information Sys-tems, 17:56-61.

Liber O, Johnson M (2008) Personal Learning Environments. Interactive Learning Environments, 16:1-2.

(27)

References 17

Lops P, de Gemmis M, Semeraro G (2011) Content-based Recommender Systems: State of the art and Trends. In: Ricci F, Rokach L, Shapira B (eds) Recommender Systems Handbook, Springer US, 73-105.

Malone T, Grant K, Turbak F, Brobst S, Cohen M (1987) Intelligent information sharing systems. Communications of the ACM, 30(5):390-402.

Manouselis N, Costopoulou C (2007) Analysis and Classification of Multi-Criteria Recommender Systems. World Wide Web: Internet and Web Information Sys-tems, Special Issue on Multi-channel Adaptive Information Systems on the World Wide Web, 10(4):415-441.

Manouselis N, Vuorikari R, Van Assche F (2007) Simulated Analysis of MAUT Collaborative Filtering for Learning Object Recommendation. In: Proceedings of the Workshop on Social Information Retrieval in Technology Enhanced Learning (SIRTEL 2007), Crete, Greece, 2007.

Manouselis N, Drachsler H, Verbert K, Santos OC (eds) (2010). Proceedings of the 1st Workshop on Recommender Systems for Technology Enhanced Learning (RecSysTEL 2010), Procedia Computer Science, (1)2.

Manouselis N, Drachsler H, Vuorikari R, Hummel H, Koper R (2011) Recom-mender Systems in Technology Enhanced Learning. In: Kantor P, Ricci F, Rokach L, Shapira B (eds), Recommender Systems Handbook, pp. 387-415, Springer US. Moore JD, Swartout WR (1990) Pointing: A way toward explanation dialogue. In

Eight National Conference on Artificial Intelligence, AAAI, pp. 457-464. Ochoa X (2011) Modeling the Macro-Behavior of Learning Object Repositories.

Interdisciplinary Journal of E-Learning and Learning Objects, 7:25-35.

Oppermann R (1994) Adaptively supported adaptability. International Journal of Human Computer Studies, 40(3):455-472.

Paris C (1988) Tailoring object description to a user’s level of expertise. Computa-tional Linguistics, 14(3):64-78.

Pazzani M, Billsus D (1997) Learning and Revising User Profiles: The Identification of Interesting Web Sites. Machine Learning, 27:313-331.

Pennock DM, Horvitz E (1999) Analysis of the axiomatic foundations of collabo-rative filtering. In: Proceedings of the AAAI Workshop on Artificial Intelligence for Electronic Commerce, Orlando, Florida, July 1999.

Prez T, Gutirrez J, Lopistguy P (1995) An adaptive hypermedia system. In: Greer J (eds) Proceedings of the 7th World Conference on Artificial Intelligence in Edu-cation, pp. 351-358. Washington, DC, AACE.

Reinhardt W, Meier C, Drachsler H, Sloep P (2011a) Analyzing 5 years of EC-TEL proceedings. In: Kloos CD, Gillet D, Garca R, Wild F, Wolpers M (eds) Towards Ubiquitous Learning Proceedings of the 6th European Conference on Technology Enhanced Learning.

Reinhardt W, Wilke A, Moi M, Drachsler H, Sloep P (2011b) Mining and visualiz-ing Research Networks usvisualiz-ing the Artefact-Actor-Network approach. In: Abraham A (ed) Learning, 1-34. Springer London.

(28)

Resnick P, Varian HR (1997) Recommender Systems. Communications of the ACM, 40(3):56-58.

Romero C, Ventura S (2007) Educational data mining: A survey from 1995 to 2005. Expert Systems with Applications, 33(1):135-146. Elsevier.

Romero C, Ventura S, Garcia E (2008) Data Mining in Course Management Sys-tems: MOODLE Case Study and Tutorial. Computers & Education, 51(1): 368-384.

Romero C, Ventura S (2010) Educational Data Mining: A Review of the State-of-the-Art. IEEE Transaction on Systems, Man, and Cybernetics, Part C: Applica-tions and Reviews, 40(6):601-618.

Santos OC, Boticario JG (eds) (in press) Recommender Systems to Support the Dynamics of Virtual Learning Communities. International Journal of Web Based Communities (IJWBC), Inderscience.

Santos OC, Boticario JG (eds) (2012) Educational Recommender Systems and Technologies: Practices and Challenges, IGI Global.

Schafer JB, Frankowski D, Herlocker J, Sen S (2007) Collaborative Filtering Sys-tems. In Brusilovsky P, Kobsa A, Neidl W. (eds) The Adaptive Web: Methods and Strategies of Web Personalization, Lecture Notes in Computer Science, Vol. 4321, pp. 291-324, Springer-Verlag, Berlin Heidelberg New York.

Schafer JB, Konstan JA, Riedl J (2001) E-Commerce Recommendation Applica-tions. Data Mining and Knowledge Discovery, 5:115-153.

Schein AI, Popescul A, Ungar L, Pennock DM (2005) CROC: A New Evaluation Criterion for Recommender Systems. Electronic Commerce Research, 5:51-74. Schneider-Hufschmidt M, Kuhme T, Malinowski U (eds) (1993) Adaptive user

interfaces: Principles and practice. Human Factors in Information Technology. North-Holland, Amsterdam.

Shardanand U, Maes P (1995) Social Information Filtering: Algorithms for Auto-matic ’Word of Mouth. In: Proceedings of the Conference on Human Factors in Computing Systems (CHI’95), Denver CO, USA.

Siemens G (2010) What are Learning Analytics? Retrieved 11 March 2012 from http://www.elearnspace.org/blog/2010/08/25/what-are-learning-analytics/ Siemens G, Gasevic D (eds) (2011) Proceedings of the 1st conference on Learning

Analytics and Knowledge, Banff, Calgary, Canada. ACM.

Smyth B (2007) Case-based Recommendation. In: Brusilovsky P, Kobsa A, Neidl W (eds) The Adaptive Web: Methods and Strategies of Web Personalization, Lec-ture Notes in Computer Science, Vol. 4321, pp. 342-376. Springer-Verlag, Berlin Heidelberg New York.

Tang TY, Daniel BK, Romero C (eds) (to appear) Recommender systems for and in social and online learning environments. Expert Systems The Journal of Knowl-edge Engineering.

van Bruggen J, Sloep P, van Rosmalen P, Brouns F, Vogten H, Koper R, Tatter-sall C (2004) Latent semantic analysis as a tool for learner positioning in learn-ing networks for lifelong learnlearn-ing. British Journal of Educational Technology, 35(6):729-738

(29)

References 19

Verbert K, Duval E, Lindstaedt SN, Gillet D (eds) (2010) Context-aware recom-mender systems. Journal of Universal Computer Science, 16(16):2175-2178. Vuorikari R, Koper R (2009) Ecology of social search for learning resources.

Campus-Wide Information Systems, 26(4):272-286.

Vuorikari R, Manouselis N, Duval E (eds) (2009) Special issue on social information retrieval for technology enhanced learning. Journal of Digital Information, 10(2). Waldrop M (1992) Complexity: The Emerging Science at the Edge of Chaos. Simon

and Schuster, New York.

Weibelzahl S (2003) Evaluation of Adaptive Systems. PhD Dissertation. University of Trier, Germany.

Wild F, Moedritscher F, Sigurdarson SE (2008) Designing for Change: Mash-Up Personal Learning Environments. eLearning Papers, 9.

Wilson S, Sharples P, Griffith D (2008) Distributing education services to personal and institutional systems using Widgets. In: Wild F, Kalz M, Palmer M (eds)

(30)

Recommender Systems for Learning

Nikos Manouselis, Hendrik Drachsler, Katrien

Verbert and Erik Duval